CSRLinear

class brainpy.dnn.CSRLinear(conn, weight, sharding=None, mode=None, name=None, method='cusparse', transpose=True)

Synaptic matrix multiplication with CSR sparse computation.

It performs the computation of:

\[y = x @ M\]

where \(y\) is the postsynaptic value, \(x\) the presynaptic value, \(M\) the synaptic weight using a CSR sparse matrix.

Parameters:

• conn (TwoEndConnector) – TwoEndConnector. The connection.

• weight (Union[float, TypeVar(ArrayType, Array, Variable, TrainVar, Array, ndarray), Callable]) – Synaptic weights. Can be a scalar, array, or callable function.

• sharding (Union[Sequence[str], Device, Sharding, None]) – The sharding strategy.

• mode (Optional[Mode]) – The synaptic computing mode.

• name (Optional[str]) – The synapse model name.

update(x)

The function to specify the updating rule.